| <?php | <?php |
| require_once '../include/common.inc.php'; | require_once '../include/common.inc.php'; |
| require($basePath . 'lib/phpquery/phpQuery/phpQuery.php'); | require($basePath.'lib/phpquery/phpQuery/phpQuery.php'); |
| $db = $server->get_db('disclosr-agencies'); | $db = $server->get_db('disclosr-agencies'); |
| $rows = $db->get_view("app", "byName")->rows; | $rows = $db->get_view("app", "byName")->rows; |
| $nametoid = Array(); | $nametoid = Array(); |
| $accounts = Array(); | $accounts = Array(); |
| foreach ($rows as $row) { | foreach ($rows as $row) { |
| $nametoid[trim($row->key)] = $row->value; | $nametoid[trim($row->key)] = $row->value; |
| } | } |
| function extractHTMLAccounts($url, $accountType) { | function extractHTMLAccounts($url, $accountType) { |
| global $accounts, $nametoid; | global $accounts, $nametoid; |
| $request = Requests::get($url); | $request = Requests::get($url); |
| $doc = phpQuery::newDocumentHTML($request->body); | $doc = phpQuery::newDocumentHTML($request->body); |
| phpQuery::selectDocument($doc); | phpQuery::selectDocument($doc); |
| foreach (pq('tr')->elements as $tr) { | foreach (pq('tr')->elements as $tr) { |
| //echo $tr->nodeValue.PHP_EOL; | //echo $tr->nodeValue.PHP_EOL; |
| $agency = ""; | $agency = ""; |
| $url = ""; | $url = ""; |
| foreach ($tr->childNodes as $td) { | foreach ($tr->childNodes as $td) { |
| $class = $td->getAttribute("class"); | $class = $td->getAttribute("class"); |
| //echo "cccc $class ".$td->nodeValue.PHP_EOL; | //echo "cccc $class ".$td->nodeValue.PHP_EOL; |
| if ($class == "s11" || $class == "s10" || $class == "s7") { | if ($class == "s11" || $class == "s10" || $class == "s7") { |
| $agency = $td->nodeValue; | $agency = $td->nodeValue; |
| } else if ($class == "s6" || $class == "s9") { | } else if ($class == "s6" || $class == "s9"){ |
| $url = $td->nodeValue; | $url = $td->nodeValue; |
| foreach ($td->childNodes as $a) { | foreach($td->childNodes as $a) { |
| $href = $a->getAttribute("href"); | $href = $a->getAttribute("href"); |
| if ($href != "") { | if ($href != "") { |
| $url = $href; | $url = $href; |
| } | |
| } | |
| } | |
| } | |
| if ($agency != "" && $url != "") { | |
| if (!in_array(trim($agency), array_keys($nametoid))) { | |
| echo trim($agency)." missing" . PHP_EOL; | |
| } else { | |
| // echo $agency." = ".$url.PHP_EOL; | |
| $accounts[$nametoid[trim($agency)]][$accountType][] = $url; | |
| } | } |
| } | |
| } | } |
| } | |
| if ($agency != "" && $url != "") { | |
| if (!in_array(trim($agency), array_keys($nametoid))) { | |
| echo trim($agency) . " missing" . PHP_EOL; | |
| } else { | |
| // echo $agency." = ".$url.PHP_EOL; | |
| $accounts[$nametoid[trim($agency)]][$accountType][] = $url; | |
| } | |
| } | |
| } | } |
| } | } |
| function extractCSVAccounts($url, $accountType, $nameField, $accountField, $filter) { | function extractCSVAccounts($url, $accountType, $nameField, $accountField, $filter) { |
| global $accounts, $nametoid; | global $accounts, $nametoid; |
| $request = Requests::get($url); | $request = Requests::get($url); |
| $Data = str_getcsv($request->body, "\n"); //parse the rows | $Data = str_getcsv($request->body, "\n"); //parse the rows |
| $headers = Array(); | $headers = Array(); |
| foreach ($Data as $num => $line) { | foreach ($Data as $num => $line) { |
| $Row = str_getcsv($line, ",", '"'); | $Row = str_getcsv($line, ",",'"'); |
| if ($num == 0) { | if ($num == 0) { |
| } else if ($num == 1) { | } else if ($num == 1) { |
| $headers = $Row; | $headers = $Row; |
| //print_r($headers); | //print_r($headers); |
| } else { | } else { |
| if (isset($Row[array_search($nameField, $headers)])) { | if (isset($Row[array_search($nameField, $headers)])) { |
| $agencyName = $Row[array_search($nameField, $headers)]; | $agencyName = $Row[array_search($nameField, $headers)]; |
| if (!$filter || $Row[array_search("State", $headers)] == "NAT") { | if (!$filter || $Row[array_search("State", $headers)] == "NAT") { |
| if (!in_array(trim($agencyName), array_keys($nametoid))) { | if (!in_array(trim($agencyName), array_keys($nametoid))) { |
| echo trim($agencyName) . " missing" . PHP_EOL; | echo trim($agencyName)." missing" . PHP_EOL; |
| } else { | } else { |
| // echo $Row[array_search($nameField, $headers)] . PHP_EOL; | // echo $Row[array_search($nameField, $headers)] . PHP_EOL; |
| $accounts[$nametoid[trim($agencyName)]][$accountType][] = $Row[array_search($accountField, $headers)]; | $accounts[$nametoid[trim($agencyName)]][$accountType][] = $Row[array_search($accountField, $headers)]; |
| } | } |
| } | } |
| } else { | } else { |
| //echo "error finding agency" . $line . PHP_EOL; | //echo "error finding agency" . $line . PHP_EOL; |
| } | } |
| } | } |
| } | } |
| } | } |
| // http://agimo.govspace.gov.au/page/gov2register/ | // http://agimo.govspace.gov.au/page/gov2register/ |
| extractCSVAccounts("https://docs.google.com/spreadsheet/pub?key=0Ap1exl80wB8OdHNKVmQ5RVlvQWpibDAxNHkzcU1nV2c&single=true&gid=0&output=csv", "Twitter", "Agency/Body/Event", "", true); | extractCSVAccounts("https://docs.google.com/spreadsheet/pub?key=0Ap1exl80wB8OdHNKVmQ5RVlvQWpibDAxNHkzcU1nV2c&single=true&gid=0&output=csv", "Twitter", "Agency/Body/Event", "", true); |
| // RSS | // RSS |
| extractHTMLAccounts("https://docs.google.com/spreadsheet/pub?hl=en_GB&hl=en_GB&key=0Ah41IAK0HzSTdGJxandJREhLSGlWWUZfZ2xKOTNHZ0E&output=html", "RSS"); | extractHTMLAccounts("https://docs.google.com/spreadsheet/pub?hl=en_GB&hl=en_GB&key=0Ah41IAK0HzSTdGJxandJREhLSGlWWUZfZ2xKOTNHZ0E&output=html", "RSS"); |
| extractHTMLAccounts("https://docs.google.com/spreadsheet/pub?hl=en_GB&hl=en_GB&key=0Ah41IAK0HzSTdGtjcW9vOXdyZ3pOV21vQU51VmhzQnc&single=true&gid=0&output=html", "Facebook"); | extractHTMLAccounts("https://docs.google.com/spreadsheet/pub?hl=en_GB&hl=en_GB&key=0Ah41IAK0HzSTdGtjcW9vOXdyZ3pOV21vQU51VmhzQnc&single=true&gid=0&output=html", "Facebook"); |
| foreach ($accounts as $id => $accountTypes) { | |
| echo $id . "<br>" . PHP_EOL; | |
| $doc = object_to_array($db->get($id)); | |
| // print_r($doc); | |
| foreach ($accountTypes as $accountType => $accounts) { | |
| if (!isset($doc["has" . $accountType]) || !is_array($doc["has" . $accountType])) { | |
| $doc["has" . $accountType] = Array(); | |
| } | |
| $doc["has" . $accountType] = array_unique(array_merge($doc["has" . $accountType], $accounts)); | |
| } | |
| $db->save($doc); | |
| } | |
| ?> | ?> |
| #http://packages.python.org/CouchDB/client.html | |
| import couchdb | |
| import json | |
| import pprint | |
| import re | |
| from tidylib import tidy_document | |
| couch = couchdb.Server('http://127.0.0.1:5984/') | |
| # select database | |
| docsdb = couch['disclosr-documents'] | |
| def f(x): | |
| invalid = re.compile(r"ensure|testing|flicker|updating|longdesc|Accessibility Checks|not recognized") | |
| valid = re.compile(r"line") | |
| return (not invalid.search(x)) and valid.search(x) and x != '' | |
| for row in docsdb.view('app/getValidationRequired'): | |
| print row.id | |
| html = docsdb.get_attachment(row.id,row.value.iterkeys().next()).read() | |
| #print html | |
| document, errors = tidy_document(html,options={'accessibility-check':1,'show-warnings':0,'markup':0},keep_doc=True) | |
| #http://www.aprompt.ca/Tidy/accessibilitychecks.html | |
| #print document | |
| errors = '\n'.join(filter(f,errors.split('\n'))) | |
| #print errors | |
| doc = docsdb.get(row.id) | |
| doc['validation'] = errors | |
| docsdb.save(doc) | |
| <?php | <?php |
| include_once('include/common.inc.php'); | include_once('include/common.inc.php'); |
| //include_header(); | //include_header(); |
| $format = "html"; | $format = "html"; |
| if (isset($_REQUEST['format'])) { | if (isset($_REQUEST['format'])) { |
| $format = $_REQUEST['format']; | $format = $_REQUEST['format']; |
| } | } |
| function add_node($id, $label, $parent="") { | function add_node($id, $label, $parent="") { |
| global $format; | global $format; |
| if ($format == "html") { | if ($format == "html") { |
| // echo "nodes[\"$id\"] = graph.newNode({label: \"$label\"});" . PHP_EOL; | // echo "nodes[\"$id\"] = graph.newNode({label: \"$label\"});" . PHP_EOL; |
| } | } |
| if ($format == "dot" && $label != "") { | if ($format == "dot" && $label != "") { |
| echo "$id [label=\"$label\"];". PHP_EOL; | echo "$id [label=\"$label\"];". PHP_EOL; |
| } | } |
| if ($format == "gexf") { | if ($format == "gexf") { |
| echo "<node id='$id' label=\"".htmlentities($label,ENT_XML1)."\" ".($parent != ""? "pid='$parent'><viz:size value='1'/>":"><viz:size value='2'/>") | echo "<node id='$id' label=\"".htmlentities($label,ENT_XML1)."\" ".($parent != ""? "pid='$parent'><viz:size value='1'/>":"><viz:size value='2'/>") |
| ."<viz:color b='".rand(0,255)."' g='".rand(0,255)."' r='".rand(0,255)."'/>" | ."<viz:color b='".rand(0,255)."' g='".rand(0,255)."' r='".rand(0,255)."'/>" |
| ."</node>". PHP_EOL; | ."</node>". PHP_EOL; |
| } | } |
| } | } |
| function add_edge($from, $to, $color) { | function add_edge($from, $to, $color) { |
| global $format; | global $format; |
| if ($format == "html") { | if ($format == "html") { |
| // echo "graph.newEdge(nodes[\"$from\"], nodes['$to'], {color: '$color'});" . PHP_EOL; | // echo "graph.newEdge(nodes[\"$from\"], nodes['$to'], {color: '$color'});" . PHP_EOL; |
| } | } |
| if ($format == "dot") { | if ($format == "dot") { |
| echo "$from -> $to ".($color != ""? "[color=$color]":"").";". PHP_EOL; | echo "$from -> $to ".($color != ""? "[color=$color]":"").";". PHP_EOL; |
| } | } |
| if ($format == "gexf") { | if ($format == "gexf") { |
| echo "<edge id='$from$to' source='$from' target='$to' />". PHP_EOL; | echo "<edge id='$from$to' source='$from' target='$to' />". PHP_EOL; |
| } | } |
| } | } |
| if ($format == "gexf") { | if ($format == "gexf") { |
| //header('Content-Type: text/xml'); | //header('Content-Type: text/xml'); |
| header('Content-Type: application/gexf+xml'); | header('Content-Type: application/gexf+xml'); |
| echo '<?xml version="1.0" encoding="UTF-8"?> | echo '<?xml version="1.0" encoding="UTF-8"?> |
| <gexf xmlns="http://www.gexf.net/1.2draft" xmlns:viz="http://www.gexf.net/1.2draft/viz" version="1.2"> | <gexf xmlns="http://www.gexf.net/1.2draft" xmlns:viz="http://www.gexf.net/1.2draft/viz" version="1.2"> |
| <meta lastmodifieddate="2009-03-20"> | <meta lastmodifieddate="2009-03-20"> |
| <creator>Gexf.net</creator> | <creator>Gexf.net</creator> |
| <description>A hello world! file</description> | <description>A hello world! file</description> |
| </meta> | </meta> |
| <graph mode="static" defaultedgetype="directed"> | <graph mode="static" defaultedgetype="directed"> |
| <nodes>'. PHP_EOL; | <nodes>'. PHP_EOL; |
| } | } |
| if ($format == "dot") { | if ($format == "dot") { |
| echo 'digraph g {'. PHP_EOL; | echo 'digraph g {'. PHP_EOL; |
| } | } |
| $db = $server->get_db('disclosr-agencies'); | $db = $server->get_db('disclosr-agencies'); |
| add_node("fedg","Federal Government - Commonwealth of Australia"); | add_node("fedg","Federal Government - Commonwealth of Australia"); |
| try { | try { |
| $rows = $db->get_view("app", "byCanonicalName", null, true)->rows; | $rows = $db->get_view("app", "byCanonicalName", null, true)->rows; |
| //print_r($rows); | //print_r($rows); |
| foreach ($rows as $row) { | foreach ($rows as $row) { |
| add_node($row->id, $row->key); | add_node($row->id, $row->key); |
| } | } |
| } catch (SetteeRestClientException $e) { | } catch (SetteeRestClientException $e) { |
| setteErrorHandler($e); | setteErrorHandler($e); |
| } | } |
| if ($format == "gexf") { | if ($format == "gexf") { |
| echo '</nodes> | echo '</nodes> |
| <edges>'. PHP_EOL; | <edges>'. PHP_EOL; |
| } | } |
| try { | try { |
| $rows = $db->get_view("app", "byDeptStateName", null, true)->rows; | $rows = $db->get_view("app", "byDeptStateName", null, true)->rows; |
| //print_r($rows); | //print_r($rows); |
| foreach ($rows as $row) { | foreach ($rows as $row) { |
| add_edge("fedg", $row->value, 'yellow'); | add_edge("fedg", $row->value, 'yellow'); |
| } | } |
| } catch (SetteeRestClientException $e) { | } catch (SetteeRestClientException $e) { |
| setteErrorHandler($e); | setteErrorHandler($e); |
| } | } |
| try { | try { |
| $rows = $db->get_view("app", "parentOrgs", null, true)->rows; | $rows = $db->get_view("app", "parentOrgs", null, true)->rows; |
| // print_r($rows); | // print_r($rows); |
| foreach ($rows as $row) { | foreach ($rows as $row) { |
| add_edge($row->key, $row->value, 'blue'); | add_edge($row->key, $row->value, 'blue'); |
| } | } |
| } catch (SetteeRestClientException $e) { | } catch (SetteeRestClientException $e) { |
| setteErrorHandler($e); | setteErrorHandler($e); |
| } | } |
| if ($format == "html") { | if ($format == "html") { |
| ?> | ?> |
| <div id="sigma-example" width="960" style="min-height:800px;background-color: #333;"></div> | <div id="sigma-example" width="960" style="min-height:800px;background-color: #333;"></div> |
| <script src="javascripts/sigma.min.js"></script> | <script src="javascripts/sigma.min.js"></script> |
| <script src="javascripts/sigma/plugins/sigma.parseGexf.js"></script> | <script src="javascripts/sigma/plugins/sigma.parseGexf.js"></script> |
| <script src="javascripts/sigma/plugins/sigma.forceatlas2.js"></script> | <script src="javascripts/sigma/plugins/sigma.forceatlas2.js"></script> |
| <script type="text/javascript">function init() { | <script type="text/javascript">function init() { |
| // Instanciate sigma.js and customize rendering : | // Instanciate sigma.js and customize rendering : |
| var sigInst = sigma.init(document.getElementById('sigma-example')).drawingProperties({ | var sigInst = sigma.init(document.getElementById('sigma-example')).drawingProperties({ |
| defaultLabelColor: '#fff', | defaultLabelColor: '#fff', |
| defaultLabelSize: 14, | defaultLabelSize: 14, |
| defaultLabelBGColor: '#fff', | defaultLabelBGColor: '#fff', |
| defaultLabelHoverColor: '#000', | defaultLabelHoverColor: '#000', |
| labelThreshold: 6, | labelThreshold: 6, |
| defaultEdgeType: 'curve' | defaultEdgeType: 'curve' |
| }).graphProperties({ | }).graphProperties({ |
| minNodeSize: 0.5, | minNodeSize: 0.5, |
| maxNodeSize: 5, | maxNodeSize: 5, |
| minEdgeSize: 5, | minEdgeSize: 5, |
| maxEdgeSize: 5 | maxEdgeSize: 5 |
| }).mouseProperties({ | }).mouseProperties({ |
| maxRatio: 32 | maxRatio: 32 |
| }); | }); |
| // Parse a GEXF encoded file to fill the graph | // Parse a GEXF encoded file to fill the graph |
| // (requires "sigma.parseGexf.js" to be included) | // (requires "sigma.parseGexf.js" to be included) |
| sigInst.parseGexf('graph.php?format=gexf'); | sigInst.parseGexf('graph.php?format=gexf'); |
| sigInst.bind('downnodes',function(event){ | sigInst.bind('downnodes',function(event){ |
| var nodes = event.content; | var nodes = event.content; |
| }); | }); |
| // Draw the graph : | |
| sigInst.draw(); | |
| // Start the ForceAtlas2 algorithm | // Start the ForceAtlas2 algorithm |
| // (requires "sigma.forceatlas2.js" to be included) | // (requires "sigma.forceatlas2.js" to be included) |
| sigInst.startForceAtlas2(); | sigInst.startForceAtlas2(); |
| // Draw the graph : | |
| sigInst.draw(); | |
| } | } |
| if (document.addEventListener) { | if (document.addEventListener) { |
| document.addEventListener("DOMContentLoaded", init, false); | document.addEventListener("DOMContentLoaded", init, false); |
| } else { | } else { |
| window.onload = init; | window.onload = init; |
| } | } |
| </script> | </script> |
| <?php | <?php |
| } | } |
| if ($format == "dot") { | if ($format == "dot") { |
| echo "}"; | echo "}"; |
| } | } |
| if ($format == "gexf") { | if ($format == "gexf") { |
| echo ' </edges> | echo ' </edges> |
| </graph> | </graph> |
| </gexf>'. PHP_EOL; | </gexf>'. PHP_EOL; |
| } | } |
| //include_footer(); | //include_footer(); |
| ?> | ?> |
| <?php | <?php |
| include $basePath . "schemas/schemas.inc.php"; | include $basePath . "schemas/schemas.inc.php"; |
| require ($basePath . 'couchdb/settee/src/settee.php'); | require ($basePath . 'couchdb/settee/src/settee.php'); |
| function createDocumentsDesignDoc() { | function createDocumentsDesignDoc() { |
| /*"views": { | /*"views": { |
| "web_server": { | "web_server": { |
| "map": "function(doc) {\n emit(doc.web_server, 1);\n}", | "map": "function(doc) {\n emit(doc.web_server, 1);\n}", |
| "reduce": "function (key, values, rereduce) {\n return sum(values);\n}" | "reduce": "function (key, values, rereduce) {\n return sum(values);\n}" |
| }, | }, |
| "byAgency": { | "byAgency": { |
| "map": "function(doc) {\n emit(doc.agencyID, 1);\n}", | "map": "function(doc) {\n emit(doc.agencyID, 1);\n}", |
| "reduce": "function (key, values, rereduce) {\n return sum(values);\n}" | "reduce": "function (key, values, rereduce) {\n return sum(values);\n}" |
| }, | }, |
| "byURL": { | "byURL": { |
| "map": "function(doc) {\n emit(doc.url, doc);\n}" | "map": "function(doc) {\n emit(doc.url, doc);\n}" |
| }, | }, |
| "agency": { | "agency": { |
| "map": "function(doc) {\n emit(doc.agencyID, doc);\n}" | "map": "function(doc) {\n emit(doc.agencyID, doc);\n}" |
| }, | }, |
| "byWebServer": { | "byWebServer": { |
| "map": "function(doc) {\n emit(doc.web_server, doc);\n}" | "map": "function(doc) {\n emit(doc.web_server, doc);\n}" |
| }, | } |
| "getValidationRequired": { | |
| "map": "function(doc) {\nif (doc.mime_type == \"text/html\" \n&& typeof(doc.validation) == \"undefined\") {\n emit(doc._id, doc._attachments);\n}\n}" | |
| } | |
| }*/ | }*/ |
| } | } |
| function createAgencyDesignDoc() { | function createAgencyDesignDoc() { |
| global $db; | global $db; |